June 4, 2024Open Access

Porting the grid-based 3D+3V hybrid-Vlasov kinetic plasma simulation Vlasiator to heterogeneous GPU architectures

View Full Paper

Ask AI

Bookmark

Authors

MBMarkus BattarbeeHelsinki Institute of Physics KPKonstantinos PapadakisUniversity of Helsinki UGUrs GanseUniversity of Helsinki

Discussion

Member takes

Implication

Simulation study demonstrates cross-platform GPU acceleration for the Vlasiator space plasma model, highlighting a pathway toward exascale kinetic calculations.

Key Points

The Vlasiator space plasma code achieves cross-platform portability on both NVIDIA and AMD architectures, preserving ion-kinetic dynamics across adaptive six-dimensional grids.
Algorithmic redesign of the Vlasov equation incorporates unified memory management, automated kernel construction, and a parallel block adjustment method for the sparse velocity mesh.
Porting to heterogeneous CPU and GPU architectures establishes a path toward multi-node exascale simulations, allowing efficient advection modeling across vast plasma domains.

Cite This Study

Battarbee et al. (2024) studied this question.

synapsesocial.com/papers/68e664c0b6db6435875f1aea https://doi.org/10.48550/arxiv.2406.02201

View Full Paper

Ask AI

Bookmark

Implication

Simulation study demonstrates cross-platform GPU acceleration for the Vlasiator space plasma model, highlighting a pathway toward exascale kinetic calculations.

Porting the grid-based 3D+3V hybrid-Vlasov kinetic plasma simulation Vlasiator to heterogeneous GPU architectures

Discussion

Member takes

Implication

Key Points

Cite This Study

Porting the grid-based 3D+3V hybrid-Vlasov kinetic plasma simulation Vlasiator to heterogeneous GPU architectures

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider